An electrical power conversion circuit is a circuit in which electrical power is changed so that a power source with a voltage or current can serve a load requiring a different voltage or current. In switch mode conversion circuits and power conversion techniques, power is typically changed from a supply that is higher in current or voltage to serve a load requiring lower current or voltage, or from a supply that is lower in voltage or current to a load requiring higher voltage or current. Where high conversion efficiency is made possible by design of the circuit, power output is substantially equal to power input. Where voltage-ampere products (VA) are close to the same at both output and input, and assuming a fixed VA at the input, a reduction in voltage at a resistive load is necessarily accompanied by an increase in load current, and vice-versa.
There are three basic topologies of switch mode power converters. They include step-down (buck--see FIG. 1), step-up (boost), inverting (including flyback) converters, and "duals" of these three. A "dual" version of any of the three basic class devices can sometimes be effected by a simple transform as follows: series inductors become parallel capacitors; and parallel capacitors becomes series inductors. Other transforms can be more involved, and the above transforms are provided as illustrative only, as will be appreciated by those skilled in the art. Dual topologies are known to have the following characteristics: a) discontinuous currents to and from the voltage sources become continuous currents to and from voltage sources; b) the DC transfer function (output voltage vs. input voltage vs. duty cycle) remains the same; and c) the dual input and output inductors can be combined together in one magnetic structure.
In all such topologies, at least with respect to DC functionality, each converter typically consists of two switches, an inductor, and input and output filters. Nearly all conventional converters are some derivation or combination of these topologies and their duals.
Output voltage regulation in these known converter topologies then is achieved by varying the duty cycle of the switches. FIG. 1 shows a simple buck converter with ideal switches. In DC conversion circuits of the type generally known as buck or boost regulators, the two solid state switches typically employed are reciprocally and cyclically operated so that one switch is "on" or conducting, while the other is "off" or non-conducting, and vice-versa. Thus as the duty cycle of the modulation of the two switches is varied, so is the voltage (or current) conversion ratio varied between source and load. For example, in a buck topology, if S1 is modulated at a given duty cycle D and S2 is modulated exactly opposite S1 (S2 closed when S1 is open and vice versa) then output voltage is given by the formula: EQU V.sub.out =V.sub.in *D.
Other known formulae similarly apply to other respective known regulator classes, as will be appreciated by those skilled in the art. This relationship holds for either polarity of V.sub.in. Theoretically then, alternating voltage on the input would manifest itself on the output according to the same relationship, assuming the use of "ideal" switches. However, as a matter of practice, in the absence of such ideal switches, conventional single stage converter implementations do not function in AC to AC conversions.
In conversion circuits used to drive reactive loads such as induction motors, bi-directional energy flows and other four quadrant operation must also be accommodated. Simple power transformers are in common usage, though necessarily restricted to AC power conversion; however, they grow heavy and bulky as power levels increase, and they are by nature not readily variable in their conversion ratios without some kind of tap changing modification.
There has been substantial work done in the area of AC converters, but known methods suffer for one or more reasons. For instance, some employ simplistic control schemes that lead to a variety of failure modes in the switches.
In all real world switches, there exist timing delays and finite rise/fall times, both of which vary from device to device and over varying operating conditions. If care is not taken in a conventional two switch converter as outlined above, both switches could conduct simultaneously, with attendant high currents and excessive power dissipation which can destroy the switches.
Power transistors of some of the types commonly employed as switches in converter topologies (and other semiconductors similarly employed) are known to store significant amounts of charge, and if a control voltage is applied to turn one transistor off as the control voltage is being applied to turn the other transistor on, the flow of current in the first transistor would continue for sometime after the turn off control, and simultaneous conduction in both transistors would occur to cause a short across the power source, with potentially damaging current flow through the switches.
A simultaneous "off" condition for both transistors is also a problem, for if the first transistor is turned off before the second transistor is turned on, the series inductor in such regulating circuits (in series with the opening switch) would discharge, or force current, through the opening switch and subject the switch to potentially damaging voltage.
One known technique for dealing with the first phenomenon is the addition of a switching delay, or dead time, into the turn-on of each switch, after turn-off of the other switch. Generally, a value for the length of the switching delay is chosen to insure that one switch is completely off before the other is enabled. But in the AC circuit supposed above, that then results in both switches being disabled at the same time. And as discussed above, if any current is flowing in the output inductor L1, then the result of both switches simultaneously disabled is a voltage spike across the switches which will likely destroy them. This spike typically has to be clamped via some snubber or clamping network, but that then results in excessive clamp power dissipation and excessive switching losses in the switches. For an example of a manifestation of this problem, and an example of this limiting solution, see U.S. Pat. No. 4,947,311 to Peterson, the disclosure of which is hereby incorporated by this reference into this disclosure as background as if fully set forth.
Another approach to the problems described above has been through the use of resonant switching circuits that employ zero voltage or current switching techniques. These are sometimes referred to as "soft switching", or zero voltage switched, techniques. Converters employing these techniques do tend to be more efficient in theory than the "hard" switching circuits using snubbers mentioned above, but some topologies have proven difficult to control, where the resonant circuit becomes increasingly unstable at lower power ranges. In addition, such resonant circuits also have more narrowly defined operating conditions (i.e., minimum and maximum current limitations), and are therefore less robust for industrial applications, and these circuits typically trade switch losses for increased conduction losses, and require bulky resonant componentry.
Switching losses into an inductive load, as encountered in conventional circuits, are generally proportional to the product of turnoff time, peak current, peak voltage, and switching frequency, and can be calculated from the well known formula: EQU P.sub.sw =0.5t.sub.off *V.sub.peak *I.sub.peak *Frequency
where P.sub.sw is the switching loss expressed in units of power. In a snubbed or clamped circuit, there is always a voltage rise across the snubber during dead time as long as either current or voltage is non-zero. As a result, there is always a significant switch or snubber dissipation. In addition, during high current surge conditions, the snubber may not be able to adequately limit the voltage rise, leading to potentially catastrophic device failure.
These known circuits therefore have significant limits in capability of conversion, at least in terms of output power, efficiency, reliability and cost. This is especially the case with present high power semiconductor technology where higher power and lower cost devices are also generally the slowest, and therefore have inherent and unacceptably high switching losses.
Recently, a variety of other topologies have been suggested for direct AC/AC converters (also sometimes simply referred to as electronic transformers). These suggested topologies have generally fallen into two types: "zero voltage switched" (ZVS) a.k.a. "soft switched"; and traditional switchmode topologies, a.k.a. "hard switched". In those topologies suggested for use in hard switched converters, it appears that the power switches are implemented either as two bidirectional switches or as four unidirectional switches. A control scheme for a two switch converter referred to above is typically some form of simple modulation (such as pulse width modulation or PWM), which schemes make use of dead times, or delays, as discussed above.
Where these suggested topologies are implemented with four switches, it appears the corresponding suggested control schemes diverge from one another. Venturini describes a scheme for switch control as a "staggered commutation", while Lipo discusses a similar method. In both of these descriptions, all four switches are required to be controlled at high speed, with both critical timing (including dead times) and level shifting required from a common controller. Cho briefly describes a different scheme which modulates two switches at high speed, but does not define any required dead times or transitions (if any are needed). Villaca employs two bidirectional, zero voltage switching (ZVS) switches operating from a simple pulse width modulator.
These suggested control schemes have a number of points in common: 1) they all have multiple switches operating simultaneously at high frequencies; 2) the timing between these switches is highly critical, in order to avoid cross conduction or voltage spikes from the output inductor; 3) maximum duty cycle may have to be limited to accommodate what may be required as fixed timing delays, and circuit response time to output overload can be consequently dangerously delayed; 4) they all apply high frequencies voltage waveforms to the output inductor, which frequencies are essentially the same from no load to full load, which results in a fixed core loss in the inductor with attendant significant power loss even at light load or no load operation, thus reducing conversion efficiency at light load with attendant increased electricity costs; 5) by employing an unchanging high frequency waveform, the switching frequency AC current component in the output inductor is also similar from no load to full load condition, so that at light load significant current is left circulating through power components like transistors, diodes, and filter capacitors, all with attendant significant power loss and reduced light load conversion efficiency and increased electricity costs; 6) they all modulate multiple switches at high frequencies, which leads to high average current requirements for control circuitry, especially with large semiconductor power devices like IGBT's MOSFET's, BJT's and MCT's.
In another vein, three phase power has been the mainstay of electrical power distribution for nearly a century, with three wire (also known as "delta") distribution being common for most situations.
Various methods are currently employed to regulate three phase power, including tap selection, magnetic synthesizers, ferroresonant transformers, inverters and the like. A circuit has been described by Mozdzer and Bose that is different from those listed above.
Other known uses for power controllers are so called static VAR compensators, which effectively add or subtract inductance or capacitance from a system, and also adaptive VAR compensators and "dynamic voltage restorers". These known devices have shortcomings in delivering a variable capacitance or variable inductance to a system, as has been reported in the literature.
AC power quality is best and electrical operating efficiency is greatest when the line current is sinusoidal in wave form and in phase with the line voltage. It is well known however that electrical elements such as reactive loads shift the line current in the mains out of phase with the line voltage. This phase shift is commonly defined in terms of "power factor", or more specifically, "displacement power factor", where displacement power factor (referred to hereafter for sake of simplicity as PF or as power factor) is given by the well known relationship: EQU PF=cos.theta.
where .theta. is the degree of phase shift "lead" (or "lag", as the case might be), also known as the phase angle, between the fundamental voltage and current. A perfect, or in-phase, relationship is equated to 1.0, while increasing degrees of phase shift are represented by power factors decreasing below unity. Power factor is also sometimes defined as the ratio of "true" power (in watts) to apparent power (in volt-amperes, or VA).
Power factor is therefore a measure of relative efficiency of power transfer and energy usage, and becomes more critical with the use of heavy draw machinery and the like, such as motors. Typical induction motor power factors can range from very low at no load to around 0.85 to 0.90 ("perfect" is 1.00) at full load. Electric utility services typically add surcharges to customers with power factors below 0.90. In addition, power factor can be highly variable, depending as it does on the instantaneous total load on the power supply. Power factor correction ("PFC") then is therefore routinely applied to compensate for bad power factor situations and, through better utilization of the existing power distribution system, to reduce the need for capital intensive additions to the power grid. Power factor correction is conventionally accomplished generally by having a series of capacitors across the line, and incrementally switching the various capacitors in and out with thyristors or relays.
Other conventional PFC techniques consist of a bridge of transistors that operate with a storage bank of capacitors or batteries (for example, see Wilkerson U.S. Pat. No. 5,283,726). A major disadvantage of this conventional PFC technique is that the output switches always switch to and from the DC storage bank voltage, which is above the peak power line voltage. As a result, switching losses are quite high.
Known automatic power factor correctors are bulky, slow and complex, and therefore only practical for large motors and groups of smaller motors. Even then, it is only the "system average" PF that is corrected, and studies show that it is preferable to correct PF at the load, rather than at the system level. For the above reasons, and other as will be appreciated by those skilled in the art, it is not practical to connect conventional PFC equipment at each load (motor). This is significant in that 60% of the electrical energy produced at this time is consumed by electric motors of 5 HP or greater in size (fully 80% of industrial consumption of electricity is for motors). And this at a time when annual electricity production is valued in excess of $52 billion! Any significant reduction in electrical consumption by these motors and/or increase in their efficiency of consumption would result in huge dollar savings.
In addition, other elements such as rectifying power supplies and SCR's, and other non-linear sources such as computers, switchmode power supplies, welders, inverters, controlled bridge rectifiers, fluorescent lights and other lights requiring a ballast, actually alter the sinusoidal waveform of the AC line current. These non-linear devices typically only draw current when the voltage is at its peak, thus causing the harmonic distortion. This is not how the power grid was designed to work.
The resulting non-sinusoidal current produced by these nonlinear loads can be mathematically resolved into a "fundamental" sine wave current at line frequency, with a number of harmonic waves at multiples of line frequency, with the fundamental producing the power in the load, while the so called "harmonics" only increase heat losses and decrease the system's power factor (that is, they generally lower the efficiency of the distribution system) with no net contribution to power in the load. EPRI (Electrical Power Research Institute) estimates predict that by the year 2010, 60% of all electrical loads will be such nonlinear, solid state electronic loads, so that the majority of all future electrical loads will consist of unwanted harmonic generators.
This is already even true for conventional PFC techniques that, in addition to other noted disadvantages, also react themselves to exacerbate harmonic problems, amplify circuit resonances, and even cause "ringing" on the mains when capacitors are switched. This ringing can cause malfunctioning and shutdowns in adjustable speed motor drives and other electronic equipment.
As discussed above, each non-linear device produces its own distorted waveform composed of varying harmonic components. Each device allows current to pass during a portion of the voltage sine wave and blocks the flow of current during another portion of the sine wave. To make matters worse, phase controlled devices such as adjustable speed drives (SCR) generate harmonic currents with amplitudes varying as a function of load change. In addition, the continued trend in government regulation appears to be towards increasing energy efficiency, and is eventually expected to mandate harmonic cleanup at the source. All of this suggests that power quality (PQ) issues are a major concern, and that there is a clear need for PFC and PQ techniques that more efficiently use power without also generating harmonic distortion themselves.
It has been suggested that the variety of power quality problems, now extant and steadily growing in magnitude and variety, are all actually summed within a power distribution system, producing effects such as: deterioration of electronic equipment performance, and continuous or sporadic computer and other microprocessor malfunctions; tripping protection circuitry of adjustable speed drives; overheating of neutral in three phase systems, leading to neutral burnout; overheating and premature failure of transformers, even when the transformer rating appears otherwise adequate; overheating of motors; nuisance tripping of circuit breakers; telephone interference; and PFC capacitor fuse blowing.
In addition, the need for improved power regulation and/or power conversion is felt in other industries as well. For instance, in the motion picture and entertainment industries, conventional light source dimming technologies produce an audible 60 Hz hum in large studio lamps. The conventional technology generally applied is dimming by means of phase angle fired triacs, which limit the energy to the lamps by basically opening the circuit to the lamps for a given percentage of every half cycle (see example of resultant broken waveform in FIG. 2). This scheme is relatively simple to implement, but is electrically very noisy, which leads to additional design problems and additional implementation costs. A means of smoothly and quietly dimming such lamps would be of great use.